0 dBFS is the digital meter's top mark for ordinary fixed-point audio files. A peak at −12 dBFS is safely below it. That spare room lets a singer or drummer play louder without hitting the recording limit; engineers call it headroom.
LESSON 22 · RECORDING & PRODUCTION · PART 2
Why do recording engineers use 24-bit and higher sample rates?
A singer hits a louder note in the take than at soundcheck. The right input level gives that note room before it turns into a crackle. The file settings affect what you can adjust later.
01 / FOLLOW THE SIGNAL
Where can a loud note get cut off?
A loud note goes through several pieces of gear before you see it on the computer. A mic or its amplifier can already sound rough by the time the recording meter shows the level. Once the take is captured cleanly, the software usually has much more room to adjust its volume.
24-bit uses finer steps than 16-bit. Both have a 0 dBFS ceiling. Read more about bit depth.
At 48 kHz, the recorder takes 48,000 readings per second. That rate sets the highest pitch it can store correctly. See how sample rate works.
02 / SET THE RECORDING LEVEL
Leave room for the next loud note
Try four settings for the same sound. The blue line shows the signal arriving at the recorder’s converter. The gold line shows what gets saved. The dashed lines mark the converter’s limits. When the signal fits, the two curves overlap. Turn off the saved line to see the blue curve by itself.
Room for a louder note
Incoming signal Saved signal
At −12 dBFS, a note 6 dB louder would still fit. This leaves room for a singer or drummer who gets louder during the take.
A quieter recording
Incoming signal Saved signal
This leaves more room for an unpredictable performance. Listen to a test take. If you turn it up later, any recorded hiss comes up along with the music.
Close to the limit
Incoming signal Saved signal
A harder drum hit or a shout could easily use up the last 3 dB. The gold line still follows the blue line in this example.
The loudest peaks get cut off
Incoming signal Saved signal
The incoming signal goes 3 dB past the limit. Look for flat spots in the gold line. That’s clipping, and it can sound rough or buzzy. Turn the saved track down and you get the same rough sound at a lower volume.
For vocals or drums, try setting the loudest rehearsal notes around −12 dBFS and listen back. That gives you room for a surprise. With 24-bit recording, the room and mic hiss usually matter more than the tiny errors from rounding each reading.
See the bit-depth calculation
In an ideal digital recording, a steady tone that reaches full scale has a signal-to-noise ratio of about 6.02 × bits + 1.76 dB. The number compares the tone with the small errors from rounding. Lower the same tone by 18 dB and its lead over those errors also drops by 18 dB.
These are calculated values for a pure test tone. An actual recording also contains noise from the room, mic and interface.
03 / FLOATING-POINT AUDIO
When can turning down a track save the take?
32-bit float is a way for software and some recording files to store very big and very small numbers. If you captured a clean vocal and turned it up too far in the software, you may be able to turn it back down. If the mic or recorder already cut off the peaks, those flat spots are part of the take.
Choose where the sound gets too loud. In each example the gold line shows the track after you turn it down by 6 dB, about half the height of the wave. Look at the shape of the tops and bottoms.
The recorder cuts off the peaks
Original sound Track after turning it down
A conventional recorder can only capture levels up to its input limit. If the signal goes past that limit, it saves flat spots where the loudest parts should be.
The flat spots remain after turning the track down. Saving that take as a 32-bit float file keeps the same shape.
The track was clean when it went into the computer
Original sound Track after turning it down
Some recording software does its internal calculations with floating-point numbers, giving it more room above the usual 0 dBFS meter ceiling. A clean signal that goes over that mark inside those calculations can often be turned down again without losing its shape. That does not rescue a microphone, preamp, converter or output that clipped earlier.
The gold peaks stay rounded. Lower the level before it reaches any effect or output that can cut off the peaks.
Some recorders can capture a wider range of levels
Original sound Track after turning it down
Some field recorders use separate converters for quieter and louder signals. That gives them room to capture a whisper and a shout in the same take. The 32-bit float file can hold the range of levels the hardware captured.
A clean peak can come back into range when you turn the file down. The mic and input circuits still have their own limits.
The mic or preamp changes the sound first
Original sound Track after turning it down
A shout into a mic or too much gain on its preamp can roughen the sound before it reaches the converter. The recorder saves the signal it receives, including that roughness.
The gold line still has flat spots after turning down the 32-bit float file.
04 / HIGHER RATES INSIDE AN EFFECT
Why can an effect add the wrong pitch?
A distortion effect changes the shape of a sound and can add pitches related to the original. Those pitches are called harmonics. Here the effect starts with a 10 kHz test tone and creates one at three times that pitch, or 30 kHz.
At a 48 kHz working rate, the highest pitch the software can represent correctly is just below 24 kHz. If an effect tries to create a pitch above that limit, it can show up as a different pitch below the limit. That’s aliasing. In the graph, follow where the second mark appears as you change the setting.
The extra tone shows up at 18 kHz
Original 10 kHz tone Unwanted 18 kHz tone
The effect tries to add a 30 kHz tone, which is 6 kHz above the 24 kHz limit. It shows up 6 kHz below the limit, at 18 kHz. Once that 18 kHz tone is there, a filter set to remove tones above 24 kHz leaves it behind.
At 96 kHz, the extra tone stays at 30 kHz
Original 10 kHz tone Added 30 kHz tone, before filtering
Working at 96 kHz raises the limit to 48 kHz, so the 30 kHz tone can be represented at its own pitch. Before saving a 48 kHz file, the software can filter out that extra tone.
Only the effect needs to work faster
Original 10 kHz tone Added 30 kHz tone inside the effect, before filtering
Some audio effects temporarily do their calculations at a higher sample rate. For example, a project at 48 kHz can let an effect work at 96 kHz, filter out unwanted very-high-frequency results, and return to 48 kHz. This is called oversampling. It can help the effect avoid creating false lower-pitched tones, without changing the whole project's file format.
The markers show where the tones would land. Whether you could hear a stray tone in a real mix depends on its pitch and level. A higher project rate also uses more storage and processing power.
Why a sound designer might record at 96 kHz
Say you want to lower a recorded sound by an octave to make an effect. That halves all its frequencies. If your mic and recorder picked up a tone at 30 kHz, it would move to 15 kHz after the pitch shift. That’s a reason to use a higher recording rate for this particular job.
05 / PRACTICAL RECORDING SETTINGS
What settings should you start with?
| What you’re doing | Starting setting | What it gives you |
|---|---|---|
| Recording a voice, instrument or band | 24-bit at 44.1 or 48 kHz | Room to set the recording level below the top of the meter. Spend time on mic position and the sound in the room. |
| Recording audio for video | 24-bit at 48 kHz | Matches the rate commonly used for video. |
| Using distortion effects or making big pitch changes | 24-bit at 48 kHz to start. Try oversampling in the effect or 96 kHz for a specific need. | Gives the effect more room to work. A higher recording rate can help if you plan to shift captured high pitches down. |
| Recording sounds with sudden level changes | A recorder designed for 32-bit float | Can handle a wider range of input levels, within what the mic and recorder can capture cleanly. |
| Sending tracks for someone else to mix | 32-bit float if their software supports it and they want that format | Keeps more room for level changes while the mix is still in progress. |
| Listening to a finished release | Use the release in a format your gear supports | A standard CD plays at 16-bit and 44.1 kHz. If you’re comparing releases, check which mix or master each one uses. |
How much space does a recording take?
These sizes are for one minute of uncompressed stereo. At the same bit depth, doubling the sample rate doubles the space it uses. A lossless file such as FLAC can save space and keep the same audio data.
The mixing lesson explores where the instruments seem to sit and why a busy song can be easy or hard to follow. Explore the listening terms imaging, separation, congestion and dynamics.
EXPLANATIONS AND SOURCES
Sources and further reading
The graphs use original, calculated examples. These sources explain how recording levels, file formats and effects work on real equipment.
- Focusrite · How high should I set interface recording gain? How much room to leave for loud notes.
- iZotope · Digital audio basics: sample rate and bit depth What the numbers on your recording settings mean.
- Sound Devices · How is a 32-bit float file recorded? How a recorder can capture both quiet and loud sounds.
- FabFilter · Pro-L oversampling Why an effect sometimes works at a higher rate.
- iZotope RX · Resampling and anti-aliasing filters Why changing to a lower sample rate involves filtering high pitches.
Editorial review on 12 November 2026.